Submission 67
Polarbear AI Pty Ltd
Submitter group: Companies.
An AI model identified the submitter group and wrote the summary, key points and recommendation summaries. Counts do not measure importance or merit.
Summary
Read the original submission for the full text.
Polarbear describes its AI study platform’s usage and argues that teaching-focused design supports learning integrity. It says its reach is concentrated in cities, AI marking makes mistakes and its survey is self-reported and non-causal. It recommends national learning-tool expectations, de-identified evidence, redesigned assessment and school access to assured tools by 2028, prioritising disadvantaged areas.
Key points
Quotes are the submitter’s own words. Text taken from documents, including scanned pages read by text-recognition software, can differ from the PDF. Check the original. Listed in the same order as on the submission’s page in the interactive guide.
Argues that students have already adopted AI for learning, so the question is what kind of tool, and that tools built to teach are used to learn.
Students have already adopted AI for learning. The question for the education system is no longer whether, but what kind. When the tool is built to teach, students use it to learn.
Argues that coaching students through their own work, anchored to official curriculum and marking criteria, makes cheating pointless.
Learning integrity is a design problem, not a detection problem. A tool that coaches a student through their own work, anchored to the official curriculum and marking criteria, makes cheating pointless.
Says its platform data show students actively engaging rather than simply copying and pasting answers.
Our students ask to be taught, not to be handed answers. Our data overwhelmingly shows that students are not simply asking a question and then copying and pasting the answer. They are actively engaging with the platform.
Argues regional students have the most to gain, citing the cost of private tutoring and its widening of the gap between families who can pay and those who cannot.
Regional students have the most to gain. About one in six Australian students receives private tutoring, at a cost that runs to thousands of dollars a year, in a sector with little or no regulation that widens the gap between families who can pay and those who cannot.
Describes grounding all platform AI flows at run time in its curated database of current study designs.
Every AI flow on our platform, whether it is explaining, setting practice or marking, is grounded at run time in our curated database of the current study designs, so the model reads the curriculum rather than remembering it.
Describes its own safeguards: student accounts and study data stored in Sydney, and no student data used to train any AI model.
student accounts and study data are stored in Sydney and all of our application services run in Google Cloud's Sydney region, and no student data is used to train any AI model.
Recommendations
Recommendations are shown beside the submission’s own words. Listed in the same order as on its page in the interactive guide.
Education Ministers should use the Australian Framework for Generative AI in Schools review and Education Services Australia to publish learning-design expectations for AI study tools.
Education Ministers, through the review of the Australian Framework for Generative AI in Schools and Education Services Australia, should publish learning-design expectations for AI study tools: teach before telling, anchor to the official curriculum and marking criteria, give teachers cohort-level visibility, and measure accuracy and publish limits. This gives effect to the Productivity Commission's December 2025 recommendation to extend Safer Technologies 4 Schools into a quality and responsible-AI assurance framework
Include teaching before telling, official curriculum and marking criteria, cohort-level teacher visibility, accuracy measurement and published limits in learning-design expectations.
Education Ministers, through the review of the Australian Framework for Generative AI in Schools and Education Services Australia, should publish learning-design expectations for AI study tools: teach before telling, anchor to the official curriculum and marking criteria, give teachers cohort-level visibility, and measure accuracy and publish limits. This gives effect to the Productivity Commission's December 2025 recommendation to extend Safer Technologies 4 Schools into a quality and responsible-AI assurance framework
Establish a national de-identified evidence base for AI in learning, held by the Australian Education Research Organisation, to which assured providers contribute cohort-level usage and outcome insights under a minimum group size.
Establish a national, de-identified evidence base for AI in learning, held by the Australian Education Research Organisation, to which assured providers contribute cohort-level usage and outcome insights under a minimum group size.
Ask ACARA and state curriculum authorities for senior assessments: supervised AI-free reasoning and method, and AI-assisted work allowing workplace-level AI use, marked on judgement, process and verification.
Ask ACARA and the state curriculum authorities to define two kinds of senior assessment. The first is AI-free and supervised: it tests first-principles reasoning and method, how a person approaches a problem they have not seen before. The second is AI-assisted: students may use AI as freely as they will in any workplace, and are marked on judgement, process and verification rather than on an output an AI could produce on its own.
Weight take-home tasks for the proposed AI-free supervised and AI-assisted assessment models, and stop relying on AI detection.
Ask ACARA and the state curriculum authorities to define two kinds of senior assessment. The first is AI-free and supervised: it tests first-principles reasoning and method, how a person approaches a problem they have not seen before. The second is AI-assisted: students may use AI as freely as they will in any workplace, and are marked on judgement, process and verification rather than on an output an AI could produce on its own. Weight take-home tasks accordingly, as NESA has begun to from 31 August 2026, and stop relying on AI detection.
Set a national goal that every senior secondary student has school access to an assured AI learning tool by 2028, prioritising regional, rural, remote and low-income schools.
Set a national goal that every senior secondary student, wherever they live, has access through their school to an assured AI learning tool by 2028, with regional, rural, remote and low-income schools first. Deliver it through existing state and Commonwealth programs, let schools choose any assured provider, and publish the effect on NAPLAN and senior certificate results by remoteness.
Deliver the school-access goal through existing state and Commonwealth programs, allow schools to choose any assured provider, and publish effects on NAPLAN and senior certificate results by remoteness.
Set a national goal that every senior secondary student, wherever they live, has access through their school to an assured AI learning tool by 2028, with regional, rural, remote and low-income schools first. Deliver it through existing state and Commonwealth programs, let schools choose any assured provider, and publish the effect on NAPLAN and senior certificate results by remoteness.